The present study extends our understanding of the development of the lateral line system of zebrafish through the postembryonic period to sexual maturity. Histologic and SEM analyses provide an opportunity for the simultaneous description of neuromast morphology, distribution and ontogeny, and the pattern of lateral line canal morphogenesis. Our results show that the general morphology and pattern of development of the cranial lateral line canals in zebrafish is similar to that reported in other teleosts (Webb, 1989a; Tarby and Webb, 2003), but that the timing of canal development relative to the growth and maturation of canal neuromasts, neuromast morphology, and the relationship of neuromast to canal morphology, in zebrafish appear to be unusual among fishes.
Pattern and Timing of Canal Development
The pattern of development of the supraorbital and mandibular canals in zebrafish is almost identical to that recently reported for the cichlid, Archocentrus nigrofasciatus (Tarby and Webb, 2003). Canal development proceeds through four stages, but the formation of soft tissue depressions (stage IIa) and grooves with ossified canal walls (stage IIb) are clearly visible in the zebrafish, whereas stage IIa could not be detected in A. nigrofasciatus (Tarby and Webb, 2003).
The timing of canal development in zebrafish clearly differs from that in Archocentrus nigrofasciatus. Canal morphogenesis in A. nigrofasciatus, is initiated in small posthatch individuals (6.0–6.5 mm SL), and neuromasts sit in grooves for just a short period of time (growth interval of 0.5–1.5 mm); canal enclosure (stage III) occurs quickly (at 6.5–11.0 mm SL) as canal neuromast length increases (in stages II–IV; Tarby and Webb, 2003). In zebrafish, neuromasts sit in the skin (stage I) for the first month of postembryonic development without any sign of canal morphogenesis. Shallow canal depressions (stage IIa) become apparent on the dorsal surface of the head (SO canal) and on the mandible (MD canal) in 10–11 mm SL individuals (∼28–30 dpf). The presumptive canal neuromasts of the supraorbital and mandibular series remain in open grooves (stage IIb) for more than a month. The delay of canal enclosure relative to initiation of neuromast elongation in zebrafish affords an opportunity to directly observe changes in the size and shape of the sensory epithelium in canal neuromasts. Furthermore, variation in the timing of the onset and duration of the different stages of canal development can provide a context in which to analyze the evolution of developmental patterns in the lateral line system among fishes.
Despite asynchrony in the development of different canal segments within and among canals, and variation in the duration of different stages of canal development, the maturation of individual canal neuromasts (e.g., initiation of elongation) in zebrafish appears to be correlated with the development of the canal segment in which it sits. At stage IIa, canal neuromasts are round, but as grooves form and canal walls ossify (stage IIb), neuromasts start to elongate transversely with a rapid increase in hair cell number. This finding provides evidence for a developmental interaction between canal neuromasts and dermal bone composing the lateral line canals, which has yet to be characterized (discussed by Hall and Hanken, 1985; Wonsettler and Webb, 1997).
Developmental Integration of Lateral Line Canals and Dermatocranial Bones
The way in which the lateral line canals become integrated into the “lateral line” bones is reported to vary among fishes (reviewed in Tarby and Webb, 2003). In some fishes, the dermatocranial bones containing the lateral line canals develop as the result of the fusion of two centers of ossification—the lateral line canal and an underlying dermal bone. A consideration of the phylogenetic distribution of those fishes in which such a “two-component” pattern of lateral line bone development suggests that there might be a phylogenetic trend from a two-component pattern in more basal teleosts (e.g., ostariophysans, including zebrafish) to a “one-component” patterns in more advanced teleosts (e.g., cichlids), where the walls of the lateral line canals extend upward from the underlying bone (discussed in Tarby and Webb, 2003). However, the results of the present study demonstrate that the SO and MD canals of zebrafish exhibit a one-component pattern of development, which is remarkably similar to that in the cichlid, Archocentrus nigrofasciatus (Tarby and Webb, 2003). There is no evidence of a two-component pattern of development in the supraorbital or mandibular canals in zebrafish. Two explanations are offered to account for this observation.
First, zebrafish may be atypical of other ostariophysans, whose lateral line bones have been reported to have a two-component pattern of development (Lekander, 1949; discussed by Tarby and Webb, 2003). The unusual neuromast morphology and the unusual association of transversely placed neuromasts in a narrow canal system in zebrafish, appear to be rare among fishes. If an alteration in the pattern of development of the lateral line canals relative to the underlying dermal bones accompanies the evolution of this specialized morphology, this may provide an explanation for the presence of a one-component pattern of lateral line bone development in zebrafish. This pattern would be particularly interesting, because the zebrafish, Danio rerio, one of several thousand members of the Family Cyprinidae (true minnows), does not appear to have any obvious specialized characteristics and could otherwise be described as a cyprinid with a “generalized” morphology.
An alternative explanation is that, while the supraorbital and mandibular canals in zebrafish demonstrate a one-component pattern of development, the other components of the cranial lateral line canal system (contained in the parietal, posttemporal, pterotic, and supratemporal bones; Cubbage and Mabee, 1996) may demonstrate a two-component pattern of development. The variation in the association of lateral line canals with particular dermal skeletal elements observed among individual zebrafish (Cubbage and Mabee, 1996) and among species (e.g., nonteleost bony fishes), suggests some degree of independence of the development of lateral line canals from the underlying dermal bones, so we speculate that this second explanation is more likely. However, a more extensive, comparative analysis of lateral line canal development among canals in zebrafish, and of homologous skeletal elements among carefully chosen species, will be necessary to evaluate this hypothesis.
Establishing Neuromast Identities in Zebrafish
In adult fishes, the identity of individual canal neuromasts can be defined by their innervation by branches of the lateral line nerves (Coombs et al., 1988; Northcutt, 1989, 1992) as well as by their position in a lateral line canal and the identity of the dermal bone with which the canal is associated (Tarby and Webb, 2003; this study). In contrast, neuromasts in fish embryos and larvae have been mapped by using fluorescent markers and have been named according to both their location and innervation (Raible and Kruse, 2000) or by inferring their location based on the placement of foramina that carry small branches of the lateral line nerve (Cubbage and Mabee, 1996). However, in embryonic and early larval stages, presumptive canal neuromasts and superficial neuromasts cannot be distinguished so this poses interesting problems for the establishment of neuromast identities.
For instance, by using SEM and histology, we identified four neuromasts in the supraorbital canal series (501–504) in zebrafish larger than 12 mm SL. Cubbage and Mabee (1996) illustrate only two pairs of “bony struts” (ossified canal walls) in the frontal bone of a 15.2 mm individual cleared and stained for ossified bone. Based on their location, we interpret these to be the canal walls associated with neuromasts SO2 and SO4; the ossified canal walls associated with SO3 are not documented. In addition, the number of nerve foramina associated with both the SO and MD canals (Cubbage and Mabee, 1996), exceeds by one the number of canal neuromasts we have identified in these canal series using SEM. We suggest that, in each case, the “extra” foramen carries a nerve branch to nearby superficial neuromasts or carries small blood vessels that supply the capillary beds that underlie neuromasts (illustrated by Jakubowski, 1963, 1967). Thus, the number of foramina in lateral line bones cannot be used to predict the number of canal neuromasts in larval zebrafish, or perhaps other larval fishes.
The mandibular neuromast M1 identified by Raible and Kruse (2000) in 4 dpf zebrafish appears not to be one of the three presumptive canal neuromasts in the MD series (our MD1–3). Instead, we interpret this to be the first of several superficial neuromasts clustered at the mandibular symphysis (see Fig. 2A). We have also demonstrated that two of the three presumptive canal neuromasts of the MD series (MD1 and 3) appear later than either MD2 or the superficial neuromasts at the mandibular symphysis. This finding is particularly interesting because it demonstrates that, in zebrafish (and perhaps in other teleosts), primary neuromasts (those that differentiate early) are not necessarily presumptive canal neuromasts and that secondary neuromasts (differentiate later) are not necessarily superficial neuromasts. Furthermore, Raible and Kruse (2000) identify three SO neuromasts in 4 dpf fish, which we interpret as presumptive canal neuromasts SO1–3. However, they do not document the presence of our SO4. Instead, they identify neuromast O1 just caudal to SO3 and indicate that it is innervated by a different nerve branch than the SO neuromasts. However, our SEM scans indicate that our SO4, which is clearly integrated into the SO canal (but is of unknown innervation) is smaller than SO1–3 and that the development of the canal segment around SO4 lags behind that of SO1–3 (Table 1). Thus, we suggest that SO4 is part of the SO series and that it differentiates later than 4 dpf and that it is not neuromast O1.
Given such discrepancies with other recent studies, we suggest that a combination of methods (e.g., SEM, histology, fluorescent markers, cleared and stained osteologic preparations) that can simultaneously track neuromast and canal morphology and nerve innervation needs to be used to establish neuromast identities at several time points through postembryonic development, as innervation patterns become more complex with the asynchronous differentiation of presumptive canal and superficial neuromasts and the enclosure of canal neuromasts in the lateral line canals.
Unusual Neuromast and Canal Morphology in Context
The elongation of neuromasts perpendicular to the axis of best physiological sensitivity of the hair cells, the absence of a population of nonsensory cells surrounding the sensory strip, and the occurrence of transverse neuromasts in what appear to be narrow lateral line canals (Webb, 1989b) are interesting and unusual morphologic attributes of zebrafish canal neuromasts, which only become apparent several weeks posthatch and have escaped notice until now.
The elongate, transverse neuromast morphology in zebrafish appears to be rare among cyprinids (for which little is known about neuromast morphology) and among teleost fishes. Transverse neuromasts are not found in the other important cyprinid model species, the goldfish (Carassius auratus, Puzdrowski, 1989; personal observation), which has neuromasts typical of narrow canal systems that are round or oval with the long axis parallel to the canal axis. However, the North American cyprinid, Notropis buccatus (= Ericymba buccata, the silverjaw minnow) has narrow, elongate, transversely placed neuromasts, which lack a prominent nonsensory population surrounding the hair cells, and are thus identical to the canal neuromasts of zebrafish (Webb and Herman, unpublished observations; contrary to illustration in Reno, 1971). The neuromasts of another group of teleost fishes, the percids (Order Perciformes), demonstrate a great deal of variation in neuromast shape, placement, and the relative sizes of the sensory and nonsensory cell populations (Jakubowski, 1967). Several percid species are described as having transverse neuromasts, and two of these have elongate oval or rectangular neuromasts with very narrow nonsensory cell populations surrounding the hair cell population (Acerina acerina and Aspro zingel, Jakubowski, 1967). A more extensive survey of neuromast morphology among fishes is clearly needed in order to put zebrafish in context.
Despite the atypical nature of the morphology of zebrafish canal neuromasts, they can be exploited as a system in which to address some fundamental developmental issues. The axis of hair cell polarization is generally parallel to the long axis of the sensory strip of canal neuromasts, which is in turn parallel to the canal axis (Rouse and Pickles, 1991a). However, our data from postembryonic zebrafish indicate that the long axis of the canal neuromasts is placed perpendicular to the axis of hair cell polarization. Thus, we suggest that the canal neuromasts of zebrafish will provide an interesting context in which to analyze the dynamics of hair cell differentiation and proliferation, especially with respect to the relationship of the axes of mitosis, cell division, and hair cell polarization and resultant changes in the shape of hair cell epithelia.
The rate of addition of hair cells in the zebrafish ear is ∼8 hair cells/day in animals ∼3.5–7 dpf (anterior and posterior maculae, calculated from data in Haddon and Lewis, 1996). The rate of addition is ∼8 hair cells/day in the utricle and sacculus, and ∼12 hair cells/day in the lagena of animals 5–30 dpf (calculated from data in Bang et al., 2001). In juvenile and adult zebrafish (up to 7 months old), the rate of hair cell addition is approximately the same (∼10 hair cells/day in the sacculus and ∼13 hair cells/day in the lagena; Higgs et al., 2003). In stark contrast, the rate of addition of hair cells in the SO canal neuromasts of zebrafish up to 3 months old is only 0.93 hair cells/day (see Fig. 9). Thus, the rate at which hair cells are added to the sensory maculae of the inner ear is 10 times the rate at which hair cells are added to the supraorbital neuromasts in zebrafish. Rates of hair cell turnover, which are only known for neuromasts in 10 dpf zebrafish (Williams and Holder, 2000), which still retain their round shape; rates of hair cell differentiation and turnover have yet to be investigated in maturing (elongating) neuromasts in postembryonic zebrafish. This slower rate of hair cell addition in neuromasts provides an opportunity to compare the dynamics of hair cell populations in two vertebrate sensory systems.
The canal neuromasts of fishes other than zebrafish generally have a prominent population of nonsensory support cells that surround the sensory strip, which is generally round or elongate, in the axis of the canal (Webb, 2000a). The size and shape of this nonsensory cell population define the considerable variation in neuromast shape found among fishes. In embryonic and early larval zebrafish, hair cells are distributed throughout the small, round neuromast (Williams and Holder, 2000; this study). As presumptive canal neuromasts elongate, hair cells increase in number, presumably differentiating from the support cells that are scattered among the hair cells and also located in a narrow ring (one to two cells) surrounding the hair cell population. Hair cells are added at the periphery of neuromasts in 10 dpf zebrafish (Williams and Holder, 2000) and, thus, likely arise from this narrow ring of support cells in older embryos and larvae. However, the proliferation of nonsensory support cells that is necessary to generate the prominent nonsensory cell population characteristic of canal neuromasts in other fishes appears not to occur in zebrafish. The absence of this large nonsensory cell population means that zebrafish canal neuromasts are more similar to the sensory epithelia of the inner ear of zebrafish than they are to the neuromasts of other fish species. This observation raises interesting questions about the temporal and spatial expression patterns of genes that determine the proportion of hair cells and support cells in sensory epithelia of both the lateral line and inner ear.
The correlation between neuromast and canal morphology has been noted in several studies. Narrow lateral line canal systems, which are described as being uniform in diameter, generally contain round or oval canal neuromasts whose major axis is parallel to the canal axis. In contrast, widened canal systems are generally non-uniform in diameter and contain neuromasts (of various shapes among species), which have a prominent transverse axis and sit in periodic constrictions along the canal (reviewed by Webb, 2000a, b). Widened canals also tend to have a membranous canal roof, which is generally ossified only in the vicinity of each canal neuromast (e.g., several percids, Jakubowski, 1967; reviewed by Webb, 1989b; Webb 2000a, b). We have demonstrated that the cranial lateral line canals in zebrafish are generally uniform in diameter, but contain elongate, transversely placed neuromasts. Besides this unusual neuromast morphology, the association of neuromasts with a prominent transverse axis with a narrow (and in most cases, well-ossified) canal system, challenges the notion that neuromast morphology is correlated with canal morphology among species (see Coombs, et al., 1988; Webb, 1989b, 2000a, b). It is not known whether this association of transverse neuromasts with a narrow canal system is a feature of cyprinid fishes other than zebrafish, because while the canal systems of cyprinids have been described in some detail (e.g., Illick, 1956; Reno, 1966; Gosline, 1974; Mayden, 1989), the morphology of the neuromasts in this diverse taxon has not been studied. However, one cyprinid (Notropis buccatus [=Ericymba buccata], Reno, 1971; Hoyt, 1972) has widened canals, of uniform diameter, which contain transverse neuromasts that are virtually identical to those of zebrafish. Thus, transverse neuromasts are found in both narrow, well ossified canals and in widened, weakly ossified canals of uniform diameter. In addition, a growing number of phylogenetically diverse taxa (e.g., Notopterus, Omarkhan, 1948; Eigenmannia, Vischer, 1989; Cataetyx, Gibbs, 1999) have been reported to have neuromasts with a prominent transverse axis, located in widened canals with a weakly ossified canal roof and a uniform diameter. The breakdown of the reported correlation between neuromast and canal morphology among fishes suggest that lateral line canal and canal neuromast morphology have evolved more independently of one another than previously thought. Analyses of the pattern and relative timing of neuromast and canal ontogeny among species with divergent adult canal morphologies, coupled with analyses of tissue interactions during neuromast maturation and canal morphogenesis, will be necessary to identify the developmental mechanisms underlying the evolutionary diversification of the lateral line system.
The zebrafish continues to provide an extraordinary amount of insight into fundamental aspects of vertebrate embryogenesis. The unusual canal neuromast morphology that we have described in postembryonic zebrafish can provide a new context for the analysis of the genetics of hair cell differentiation and hair cell polarization, and the dynamics of growth in hair cell epithelia. The accessibility of the hair cells of the lateral line system, which are found in small, discrete populations on the external surface of the head for several weeks, can provide interesting opportunities for the imaging of developing sensory epithelia and will provide a useful comparison with developmental patterns and processes in the inner ear. The development of lateral line canals and their relationship to underlying dermal bones will likely provide an interesting context for the analysis of tissue interactions in the development of the dermatocranium, especially when placed in a comparative context. So, despite the irony that the postembryonic lateral line system of this model species appears not be typical of teleost fishes (see Bolker, 1995; Bolker and Raff, 1997; Metscher and Alhberg, 1999), this species will continue to provide important insights because it has broadened the morphologic context in which patterns and mechanisms of development and evolution in the lateral line system and inner ear may be interpreted.